Exploration of conversions

Overall graph

The overall distribution of conversions, by location and by transitivity

We define the filters that we found for call sites:

And some helper traversal functions:

Modify the dataset to count conversions:

Output the graph

First we filter the dataset

And finally, we can produce the graph

Now let’s plot the non-transitive conversion distribution. First, we filter out the transitive calls:

Then, we get the frequency counts, and we print it:

This is not extremely useful, unfortunately.

We can envision another graph, with the same data but the axes changed:

Now, if we don’t mush the projects together, maybe we can gather some metadata. First, load the project stats.

Then, add LOC information to the nontransitive conversions:

Delving deeper into slick

signature	occurrences
(slick.ast.FunctionSymbol[]) => slick.lifted.FunctionSymbolExtensionMethods[]	31
(slick.ast.Type[]) => slick.ast.TypeUtil[]	61
(slick.ast.Node[]) => slick.ast.NodeOps[]	121
That was with call sites, let’s go with